Hi,

I get an

ERROR: LoadError: UndefVarError: Downloads not defined
Stacktrace:
 [1] top-level scope
   @ o:\Julia\makie.jl:604
in expression starting at o:\Julia\makie.jl:604

Commented out.

And on execution for all demo functions:

ERROR: UndefVarError: Options not defined
Stacktrace:
 [1] custom_plot()
   @ Main o:\Julia\makie.jl:16
 [2] top-level scope
   @ REPL[12]:1

Commented out.

On execution,

custom_plot()

no error, no plot appears.

This is a spinoff of the discussion in #84

We should add somewhere a section "About the Authors" with a picture and a brief bio.

It helps to establish a nice connection of the reader to the book and the authors.

Guys,
probably is too early, but I have some material using Flux that I would like to start putting up. Not revision necessary yet, just starting material (more like notes), but it will be definitely breaking changes. Could we have a different repo within the organisation for that?

As in the web version, where its written:

DataFrame(σ = ["a", "a", "a"], δ = [π, π/2, π/3])

The PDF version reads:

DataFrameσ( = ["a", "a", "a"], δ = π[, π/2, π/3])

page 69 Chapter 4 DataFrames.jl

In the past, I've been verbal about the fact that Revise.jl isn't promoted enough. So, I should write a bit about the REPL workflow and some pointers to Pluto.jl etc.

Fix two spelling confusions:

• mixmatch to mix and match according to en.wiktionary.org/wiki/mixmatch
• usecases to use cases according to en.wikipedia.org/wiki/Use_case

Thanks @rikhuijzer!

• Package.toml
• Config.toml
• metadata.yaml
• JuliaDataScience org
• README.md
• index.md
• Prepare his Acknowledgements
• Add him to the Google Analytics

Here some thoughts on this section.

The first sentence of this section shows the word "fast" in bold. Speed is also the first thing I think of in context of performance. Unfortunately there are "only" shown comparisons of memory allocations. Of course this is also an important point (when data are that big that there is not enough memory available to hold it), but even more importantly would be speeeeeeeeed.

Is there a particular reason why you "only" show @allocated and not (also) @time and/or @benchmark from BenchmarkTools.jl (didn't try that package so far)?

A test of using @time instead of @allocated for the given examples showed that there was little to no improvement or even the "better" code (needing less memory) took longer. Maybe this is (only) due to the very small sample data used in the examples.?

From VBA (that is the only language I know a little bit (if you consider that being a programming language)) I heard that it could even be possible that using "non-optimal" types can take longer, although they allocates less memory. To be concrete, that is said to be the case when using Integer compared to Long (corresponding to Int16 and Int32 in Julia). Long should be most efficient, because (modern) processors on current platforms perform operations in that precision. Thus, internally each other type is first converted to the "optimal" type and after the operations back to the input type. When this should be true it is obvious that the two additional conversion steps would take extra time.

From what I have tested so far about this topic in VBA, on my computer I couldn't really find a (big/significant) difference between the calculation times. Any insides regarding that topic in Julia?

In response to the comment in #29: I think that it is important enough to introduce in the intro.

For example, to calculate the score per group for various columns.
To make the example come alive, maybe come up with something about fields with plants.
Note that the dot in the combine is a bit tricky to figure out.

julia> df = DataFrame(group = [:A, :A, :B, :B], X = [1, 2, 3, 4], Y = [5, 6, 7, 8])
4×3 DataFrame
 Row │ group   X      Y
     │ Symbol  Int64  Int64
─────┼──────────────────────
   1 │ A           1      5
   2 │ A           2      6
   3 │ B           3      7
   4 │ B           4      8

julia> gdf = groupby(df, :group)
GroupedDataFrame with 2 groups based on key: group
First Group (2 rows): group = :A
 Row │ group   X      Y
     │ Symbol  Int64  Int64
─────┼──────────────────────
   1 │ A           1      5
   2 │ A           2      6
⋮
Last Group (2 rows): group = :B
 Row │ group   X      Y
     │ Symbol  Int64  Int64
─────┼──────────────────────
   1 │ B           3      7
   2 │ B           4      8

julia> combine(gdf, [:X, :Y] .=> mean; renamecols=false)
2×3 DataFrame
 Row │ group   X        Y
     │ Symbol  Float64  Float64
─────┼──────────────────────────
   1 │ A           1.5      5.5
   2 │ B           3.5      7.5

I think the book should have justified text, right now the text is left-aligned. See pic below for a clarification:

As I already stated in e.g. e8e1d11 it would nice to make some sections unnumbered when there is no second section on that level.

When I have seen this correct, this should be possible by appending the section entry by {-} as e.g. can be seen in

So if you consider it would be nice to have that I'll redo the suggestions in a new PR. Especially if it is that easily doable.

PS:
I am not sure if my newest comments in #217 have been noticed by one of you, since I have added them after the PR was merged. Thank you for your comments!

It might be good idea to add a packages index at the back of the book. For example, that way, people who quickly want to know how CSV.jl works can look in the index and jump to the right page.

Please do not reopen

• Filter doesn't "remove rows" only filter! does
• Introduce column selectors and ?select.
• CategoricalArrays.jl should have a section for itself.
• transform(df, ...; renamecols=false)
• example of both filter and subset with multiple conditions, as in:

   filter(row -> row.col1 >= something1 && row.col2 <= something2, df)
   # and:
   subset(df, :col1 => ByRow(>=(something1)), :col2 => ByRow(<=(something2)))

• cover vector of Symbols as a col selector in the dataframes_select.md

in the file why_julia.md, line 250, there is an inversion between Judea and Noam

That was not what judea as a Linguist type would say to noam, a ComputerScientist type.

Judea is the computer scientist and Noam the linguist. Moreover according to the line before it is Noam who speak to Judea, so I guess the line should be:

That was not what noam as a Linguist type would say to judea, a ComputerScientist type.

We need a new meme for "Data, Data Everywhere"

A final chapter with basic statistics using both DataFrames.jl and Makie.jl. Each concept we should give the intuition behind it and also get into mathematical details.

We should cover:

• Statistics: Quick Intro and Importance
• Central Tendencies Measures: mean, median and mode
• Dispersion Measures: standard deviation, variance, mean absolute deviation, quartile and IQR
• Dependence Measures: Correlation and Covariance
• Distributions
  • Normal vs Non-Normal
  • Discrete vs Continuous
  • PDF
  • CDF
• Statistical Visualizations
  • Box plots
  • Histograms
  • Density plots
  • Anscombe Quartet

Notation discussion points from #20:

1. Always using : before the start of a code block.
2. Mentioning functions like DataFrame as DataFrame() or DataFrame(...).
3. My suggestion: only Julia objects between backticks and filenames and extension names between quotation marks (like Julia's strings).

From the webpage:

But defining a function with a keyword argument without default values is no problem:

julia> f(x; y) = x^2+y
f (generic function with 1 method)

julia> f(2, y=10)
14

Am I misunderstanding, or is the statement from the book simply wrong?

The lines

don't give the same result as the previous example ... (where :id is not shown)

I think you need to rephrase the text.

Hello there!

In the 4th chapter we find the comment bellow

NOTE: This works, but there is one thing that we need to change straight away. In this example, we defined the variables name, grade_2020 and df in global scope. This means that these variables can be accessed and edited from anywhere. If we would continue writing the book like this, we would have a few hundred variables at the end of the book even though the data that we put into the variable name should only be accessed via DataFrame! The variables name and grade_2020 where never meant to be kept for long! Now, imagine that we would change the contents of grade_2020 a few times in this book. Given only the book as PDF, it would be near impossible to figure out the contents of the variable by the end. We can solve this very easily by using functions.

Is that really necessary? I think its way more readable to let what is being explained outside a function. I think it would be better for each chapter to have a single Jupyter Notebook. For an example, Ch4_dataframes.ipynb, where every variable would be constrained in this chapter, avoiding the variable to be accessed anywhere.

Hi,

This is a great introduction book! The Makie part is also the first systematic introduction to the plotting library I've seen so far.

I have a question about the usage of limits! in the examples. I've seen many places where the limits are set without the axis type, e.g. the code for Figure 8

fig, ax, pltobj = scatter(xyvals[:, 1], xyvals[:, 2]; color=xyvals[:, 3],
    label="Bubbles", colormap=:plasma, markersize=15 * abs.(xyvals[:, 3]),
    figure=(; resolution=(600, 400)), axis=(; aspect=DataAspect()))
limits!(-3, 3, -3, 3)
Legend(fig[1, 2], ax, valign=:top)
Colorbar(fig[1, 2], pltobj, height=Relative(3 / 4))

When I tried to run the code locally, I got

julia> limits!(-3., 3., -3., 3.)
ERROR: MethodError: no method matching limits!(::Float64, ::Float64, ::Float64, ::Float64)

and it hinted to me to add the axis type as the 1st argument as

limits!(ax, -3, 3, -3, 3)

I used GLMakie but not CairoMakie for running the code. Is this due to the library version I'm using? GLMakie v0.1.30

Hi there! I enjoyed reading the PDF of this book, and I think you've done a great job, and it looks impressive.

As I read through, I picked up some typos and grammatical awkwardnesses. Would you prefer me to list them separately in an issue, or make a PR to the source Markdown files? (I don't know whether you can selectively accept and reject various edits in a PR, so I'm not sure what's best...)

• Allocations: functions with a shebang ! versus without (e.g., filter! versus filter)
• Copying vs not copying: df[!, col] versus df[:, col] (link)
• CSV.jl reading a lot of .csv files with the new version 0.9 and passing a vector of Strings, details here
• CSV.File versus CSV.read. Details in the index of the CSV.jl documentation
• CategoricalArrays.jl compression with compress=true

From some valuable suggestions by Evgeny Pogrebnyak on Twitter (https://twitter.com/PogrebnyakE/status/1435707529137463302). Evgeny argues that a book title "Julia Data Science" should contain more than just plotting and dataframes transformations. We need to clarify to readers that for version 2, we plan to add:

• Statistics (for example, GLM.jl and HypothesisTests.jl)
• AlgebraOfGraphics.jl
• Machine Learning (MLJ.jl)
• maybe other things?

Hi. @rikhuijzer .

I found text there can not match up with the following code.

More specifically, you said all of their grades in the text , but you used boolean OR operator || in the code.
My suggestion is the least change, namely all -> any.

Hey folks, you all should add a big banner on the book page asking folks to leave a review on Amazon. This will help increase the visibility of the book significantly!

It should be displaying something different for the HTML and PDF instead of instructions to the reader on what to do.

Details at https://discourse.julialang.org/t/corrections-for-juliadatascience-book/69204.

• Remove the definition of a $P(A)$ as the "set"
• Fix Kolmogorov axioms
• Double-check the rest

Just something we should have in mind.

• Official Julia here: https://julialang.org/learning/books/
• Extra-Official here: https://github.com/svaksha/Julia.jl/blob/master/Resources.md#books
• Make an announcement at Discord [ANN] Books.jl and JuliaDataScience
• HackerNews

From what I have learned so far there is no "anonymous function" in

When I am correct you declare a function that is then used in filter. The => should be the Pair operator while -> is the operator to define an anonymous function, right?

If I am correct, please rephrase the text or adapt the example code.
If I am wrong, shame on me and I am not a good "student" and sorry for the noise.

We need to:

• Normalize functions and syntax to blue codestyle, also accordingly to DataFrames, Plots, Makie, AoG docs
• Create a PDF with syntax highlighting (#15)
• Create an Author Profile at Amazon.com (both @storopoli, @rikhuijzer and @lazarusA )
• Get a ISBN (Amazon Provides a free one)
• Adapt the book to Amazon.com format
• Create cover (#148)
• Write a summary
• Tag a "first edition" release on this repository

There is a link broken in Chapter 7 datavisMakie.md:

In the "See Makie’s documentation for more." It redirects to http://makie.juliaplots.org/stable/backends_and_output.html#Backends-and-Output which is broken.

cc @lazarusA

On page 165:

On page 184:

This probably requires a fix in the PDF template.

Make a README.md for the first edition

• Add some text from juliadatascience.io homepage without copy-pasting (automated, fool-proof thing)
• Book cover
• Link to Amazon.com Book
• Cheatsheets that @lazarusA created (link the README to the image to the book)

Any plans for something on https://github.com/JuliaPlots/AlgebraOfGraphics.jl ?

We should provide citation information in index.md and also on README.md with biblatex and APA (psychology, social sciences etc.), IEEE (computer science).

Here is an opinionated version. Feel free to criticize:

Subsections for Plots

Brief overview of the JuliaPlots ecosystem

• JuliaPlots Organization
• Plots.jl (what is it good for and what are its limitations)
• Makie.jl (what is it good for and what are its limitations)
• AlgebraOfGraphics.jl (what is it good for and what are its limitations) (version 2.0)

1. What is Plots.jl?

   • plot vs plot!
   • input data
   • Series Types and functions(e.g. line, line!, heatmap, heatmap!)
   • How to save a plot

2. Attributes

   • Overview (What are attributes, and the whole symbol e.g. :xticks system)
   • Series attributes
   • Plot Attributes
   • Footnote about the extra_kwargs stuff
   • Examples using the most common things that you want to do in a visualization. This could be inserted right after you introduce a specific attribute.

3. Color and Palettes
   I think we should cover colorbrewer and the ones from matplotlib (inferno, viridis, magma).
   We should cover some stuff from the Claus Wilke Fundamentals of Data Visualization book (Chapters 4 and 19). Also we should cover the three types of color usage:

   1. sequential: continuous stuff, e.g. :blues (only blue)
   2. diverging: continuous stuff, e.g. :RdBu (from red to blue)
   3. distinguishable: discrete stuff, e.g. :Set1_5

   I have a very strong positive bias towards colorbrewer Sets (e.g. palette=:Set1_5).

   We should also mention that the reader should use a colorblinded-friendly palette or colors. Maybe we should include an official statistics regarding the prevalence of any sort of colorblindness or color difficulties in the population. I remember seeing somewhere that it was around 5% of people.

4. Layouts

   • Overview on several ways to do layouts
   • the layout argument, also cover the grid
   • the @layout macro
   • specific measures with the Plots.PlotMeasures submodule
   • adding subplots incremententally. Define p1, p2, p3; then do a plot(p1, p2, p3; layout=l)

Section 1.1.1 is still little off.

No typo, check here: https://github.com/JuliaDataScience/JuliaDataScience/blob/main/contents/why_julia.md

And them check here: https://github.com/JuliaDataScience/JuliaDataScience/blob/main/contents/preface.md

We need to include leftjoin!.

It will be added in 1.3 release of DataFrames.jl and already is in main branch.

EDIT: We only cover non-allocating functions (the bang ! ones) in performance:

• Performance Section

Source code here: https://github.com/JuliaData/DataFrames.jl/blob/main/src/join/inplace.jl

PS: Thanks for the warning @bkamins!

See #110.

It should be on the front page and somewhere in the PDF which is hidden for the Amazon version.

It should not be in the Amazon book.

Add DataFrames regex. From Kevin Bonham on Slack:

I've done this ❤️ before, but Regex column selection in DataFrames... I'm just doing some data wrangling on a table with hundreds of columns with really stupid and inconsistent names... being able to do df[!, r"pattern"i] to just ignore different capitalization and find multiple things saves SO much time

These lines of code work. You can embed a video in markdown (html). if one uncomments these lines, then it will work locally. But, it will fail in CI and the pdf generation.
Ideally, it will be good to allow the animation online, and print the first frame into the pdf version(probably too much to ask).

In #148 (comment), cormullion suggested

In #148, this was partially implemented. To implement it further, the font could be changed to Barlow (#148 (comment)) which involves ensuring that the font is available locally and in CI and a solution is required for properly placing text. I personally think that Makie isn't the best tool for text placement, but we'll might get it right with lots of fiddling.

We need a logo I will talk to someone who can do that for me at UNINOVE. Any thoughts @rikhuijzer ? We should move anything stats/Bayesian so to not confuse with future endeavors.

Maybe something with Tabular Data or Line Plots. We should definitely use Julia colors.

• Update JuliaDataScience GitHub Organization Logo
• Update JuliaDataScience Book favicon site icon